🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | log inspection situation still under development | For MoarVM see #moarvm
Set by lizmat on 22 May 2021.
00:00 reportable6 left, reportable6 joined 00:04 MasterDuke left 00:12 japhb left 00:13 japhb joined 01:13 nativecallable6 left, reportable6 left, quotable6 left, bloatable6 left, squashable6 left, greppable6 left, bisectable6 left, coverable6 left, evalable6 left, statisfiable6 left, unicodable6 left, sourceable6 left, notable6 left, committable6 left, linkable6 left, shareable6 left, benchable6 left, releasable6 left, squashable6 joined, committable6 joined 01:14 bloatable6 joined, evalable6 joined, greppable6 joined, benchable6 joined 01:15 bisectable6 joined, reportable6 joined, unicodable6 joined, releasable6 joined 01:16 shareable6 joined 02:14 nativecallable6 joined 02:15 statisfiable6 joined 02:16 linkable6 joined 03:02 nine left, nine joined 03:14 coverable6 joined, quotable6 joined 03:15 notable6 joined 04:15 sourceable6 joined 06:00 reportable6 left 07:55 frost joined 08:02 frost left 08:34 MasterDuke joined
lizmat Files=1351, Tests=117115, 301 wallclock secs (36.15 usr 9.93 sys + 4175.40 cusr 336.99 csys = 4558.47 CPU) 08:40
jdv mornin 09:35
MasterDuke jdv: i just merged and bumped a nativecall fix last night. are you able to do another blin run before the release? 09:41
jdv sure 09:50
not that it matters much but whoever manages releasable6 might want to look into why it thinks github is down. 09:52
MasterDuke huh, github was down yesterday, but then came back up 09:53
09:59 sena_kun left 10:00 reportable6 joined 10:01 sena_kun joined
jdv uh, looks like its more serious - blin can't get a build... 10:02
gist.github.com/jdv/680a5c337f45af...60b3737989 10:04
who manages the whateverable stuff?
10:05 MasterDuke left 10:26 frost joined 10:27 Xliff joined 10:29 MasterDuke joined
jdv i guess the release is "on hold" until a blin run can happen 10:29
lizmat :(
jdv its always something:) 10:30
MasterDuke i think i saw something about 'no space left on device' in a recent releaseable backtrace 10:33
AlexDaniel: ping
10:36 frost left
lizmat nine ugexe is there a reason why we just shouldn't install a module, then test, and uninstall again if test failed ? 11:10
11:31 ggoebel joined
ugexe for one because it changes runtime behavior of existing code that may get run 11:38
installing a module may do that either way yea, but unavoidable. but the uninstall after makes figuring out what blew up your production systems pretty difficul 11:42
it also implies `zef test .` would need to install all your modules to test your code in a way that is consistent with `zef install .` 11:43
Geth Test-Stream/master: 6 commits pushed by (JJ Merelo)++ 11:44
ugexe CURS exists to solve these problems, but doesnt work for non default repos (and maybe only site at all)
lizmat but we don't use that yet in zef, right ? 11:45
ugexe i've had a branch using CURS for maybe a year now. but as i've said it only really works for one very specific workflow, and that workflow is different enough for the workflow used for everything else that it wouldn't make sense to implement both and pick one conditionally 11:47
lizmat question: is this statement correct: before installing a distribution, we do not know the internal dependencies between the use targets of that distribution. *after* pre-compilation we *do* know, but all internal references are just sha's that cannot be traced back (easily) to the original use targets ? 11:48
ugexe there can also be runtime dependencies 11:49
lizmat well, I guess .install not doing pre-compilation, would turn everything into runtime dependencies, would it not ? 11:50
11:51 Altai-man joined
ugexe not sure, because the `use` statements are still occuring before all `require ::($foo)` 11:52
lizmat Hmmm... how about "zef test" without any additional parameters, run the tests without "-I" statement ? 11:55
could you just move binary files from a CURS to a CURI ? 11:59
ugexe it'd have to be a bit more complex to be generally useful... like running the tests while hiding the lib/ folder
lizmat would the shas still match ?
ugexe since because will `use lib "lib"`
12:00 reportable6 left
lizmat well, people should not use any "lib" inside a test file 12:00
ugexe i agree, and doing what i said would expose that
lizmat unless it is to be able to load other modules with test functionality
ugexe if we don't hide lib/ then the test itself is adding back in a CURFS that will prevent the CURI code from loading 12:01
lizmat right
12:02 reportable6 joined
ugexe sure. but technically CURS does the same thing, but better 12:02
lizmat and "use lib" is a pragma
ugexe s/technically/theoretically/
lizmat well theoretically, atm it is just a small shim around CURI is it not ? 12:03
and afaics, the short-id of CURS is part of the sha, so any sha would not be usable inside a CURI 12:04
ugexe all its doing is extending whatever its target repo is to another location (so new modules can live in a different path but be treated as being in e.g. site)
lizmat right
ugexe it doesn't need to be much more 12:08
12:08 Altai-man left 12:11 Altai-man joined
lizmat hmmm... if Staging could provide a map of use-target -> sha it created, wouldn't we be able to create a Staging sha -> Installation sha mapping ? 12:11
nine The SHAs are the same. Otherwise this would make the whole point of Staging moot. It is explicitly intended so you can take the generated files and drop them into a CURI and everything just works. 12:15
Because that's what RPM packaging a Raku module does: install into a temporary Staging location, zip up the files and on installation of the RPM package just extract them onto the file system
ugexe: in which use cases does it not work? 12:17
ugexe installing to non default locations
nine Well it can only work for named repos (home, site, vendor, core), so much is clear
I wonder....could we give non-default location repos a name? 12:18
Even a generated one like custom:/path/to/repo 12:19
ugexe you must mean nqp::sha1(custom:path/to/repo) 12:20
because why not
nine Embrace the SHA!
ugexe but yeah i *think* i tried doing exactly that the one time i tried to get CURS to work with non named repos 12:21
i dont remember learning if it would actually work or not though
although its name would still need to match for the repo its shadowing 12:28
or so i think 12:29
nine Of course, but zef knows what repo it's shadowing and can generate the appropriate name 12:30
Or even ask the repo directly
After all the name is a public property of the CURI
ugexe github.com/ugexe/zef/commit/a35016...9R957-R961 12:31
nine exactly 12:32
13:07 ggoebel left 15:01 ggoebel joined 15:11 Util left 15:24 discord-raku-bot left 15:25 discord-raku-bot joined 15:36 ggoebel left
jdv at this point it's approaching the end of my availability window to get a release done today:( 15:43
next window for me is sunday so aiming for that now. 15:45
15:47 ggoebel joined
MasterDuke unfortunately i can't find my login info for his server, so we'll have to wait until AlexDaniel can take a look 15:52
16:21 ggoebel left, ggoebel joined
Geth Test-Stream: a5b2a2d62b | (JJ Merelo)++ | t/400-tap-formatter.t
Eliminated tests until we understand what they do
17:15 Kaiepi left
AlexDaniel Just ran out of space on the server for a brief moment, all back to normal now I hope 17:49
Ping me if any bots are unhappy 17:50
lizmat AlexDaniel++ 17:51
Altai-man AlexDaniel++ 17:53
AlexDaniel Ahh, I didn't see MasterDuke's messages because, well ... my server was out of space 17:58
And I also can't read them because colabti is not recording the logs anymore? 17:59
Ok, how did you know what was the issue, which bot complained?
.tell lizmat test 18:00
.seen lizmat
lizmat .
AlexDaniel Hmmm
18:02 Kaiepi joined
AlexDaniel Ugh 😩 18:04
Tellable ate it's own files again
That's not good
Ok good luck to me to restore that... :'( 18:05
Basically, it runs out of space, tries to write the files, writes them partially and runs out of space 18:07
There's a reason it was called FootgunDB I guess
OK I need to run now, later tonight I'll see what's the closest backup
18:08 reportable6 left 18:10 Altai-man left, reportable6 joined
Geth Test-Stream/master: 4 commits pushed by (JJ Merelo)++ 19:24
Test-Stream: e39dab7ea8 | (JJ Merelo)++ | META6.json
Fixes :poop:
Test-Stream: a2d0374304 | (JJ Merelo)++ | README.md
Badge up
20:32 Util joined 21:25 discord-raku-bot left, discord-raku-bot joined